[Pgpool-general] is there a way out of degeneration mode

[David Boreham]

>Ok I think I found the cause. Waiting for replies from the secondary
>should be only done in replication mode, not by master/slave mode.
>Please try following patches.
>
>*** pool_process_query.c	5 Jan 2006 13:08:44 -0000	1.10
>--- pool_process_query.c	24 Jan 2006 13:28:14 -0000
>***************
>*** 315,321 ****
>  		}
>  
>  		/* this is the synchronous point */
>! 		if (DUAL_MODE)
>  		{
>  			if (kind == 0)
>  			{
>--- 315,321 ----
>  		}
>  
>  		/* this is the synchronous point */
>! 		if (REPLICATION)
>  		{
>  			if (kind == 0)
>  			{
>
>  
>
Ha ! I have been staring at exactly the same bug for the past two days ;)
(I was beginning to wonder if 'master/slave' mode was really supported 
at all,
the more I read the code...)
But, there are two places in the code where that test is done, shouldn't 
your
fix be applied in both places ? (lines 273 and 318 in the current cvs 
version
of the code). Perhaps there's some logic that I don't understand that makes
the first one benign...